Accuracy of genetic code translation by transfer RNAs on the messenger RNA programmed ribosome

Misreading of messenger RNAs (mRNAs) by transfer RNAs (tRNAs) on bacterial ribosomes has been intensely discussed over the last fifty years, but quantitative information of the missense error spectrum remains scarce. It is known that aminoacyl-tRNAs (aa-tRNAs) enter the ribosome in “ternary complex” (T3) with a protein factor, the GTPase EF-Tu. Depending on if the anticodon base triplet of the tRNA matches the codon base triplet a ternary complex is accepted for GTP hydrolysis on EF-Tu or discarded in the initial selection step of mRNA reading. After initial selection non cognate tRNAs which have “survived” the initial selection step is discarded with high probability in a proofreading step, occurring after GTP hydrolysis (Thompson and Stone, 1977; Ruusala et al., 1982). It is the chemical potential difference between GTP and its hydrolytic products that makes proofreading possible and determines its universal accuracy limits (Ehrenberg and Blomberg, 1980). There has been scientific consensus that genetic code translation errors predominantly come from mRNA reading on the ribosome (0.0005 per codon) and to a lesser extent from transcription of DNA to mRNA (0.0001 per base) and aminoacylation of tRNA to aa-tRNA (0.00001 per reaction).

We have shown by biochemical experiments involving seven tRNAs reading their cognate and all near-cognate codons that initial selection of tRNAs spans over three orders of magnitude from an intrinsic accuracy of about 100 to 100 000 (Zhang et al., 2015). The initial selection accuracy is generally very high (>10 000), but there are a small number of initial error hot-spots involving middle codon position misreading by tRNAGlu and third codon position misreading by tRNAHis. Very recently, we determined the total accuracy (A=IF) for three of these tRNAs and how it is partitioned between proofreading (F) and initial (I) selection (Submitted manuscript). An amazing finding is that in the high accuracy range log(F) is a linear function of log(I) with a slope close to two. In the low accuracy range, in contrast, log(F) is constant as log(I) decreases further. Since the same discrimination parameters are at work in both proofreading and initial selection it may seem paradoxical that proofreading can remain at a high level and thus neutralize error hot spots while initial selection plummets. To explain this, we propose that the expressed fraction of the intrinsic accuracy in proofreading selection increases and compensates for the reduction of intrinsic accuracy itself, as revealed by decreasing initial selection. We point out that the slope close to two in the log(F) versus log(I) plot most naturally arises from two consecutive proofreading steps (Ehrenberg and Blomberg, 1980) rather than one, as one has hitherto postulated.

We have, finally, directly proved the existence of two consecutive proofreading steps by engineering tRNAs with different affinities to EF-Tu and studying how the proofreading selection varies with varying aa-tRNA-EF-Tu affinity (Submitted manuscript).

Thompson RC, Stone PJ (1977) PNAS 1: 198-202
Ruusala T, Ehrenberg M, Kurland CG (1982) EMBO J 1:741-745
Ehrenberg M, Blomberg C (1980) Biophys. J 31: 333-358
Zhang J, Ieong KW, Johansson M, Ehrenberg M (2015) PNAS 112 9602-9607